Current Projects / Aktuelle Projekte

Project: Probabilistic pre- and post-processing of meteorological and hydrological ensemble forecasts in the Rhine basin"

Funding Period: 2024-2026

Client: Federal Office for Hydrology, Ministry of Transport

Status: Active

The aim of this project is to investigate whether the  Model Conditional Processor-MCP a method to process the uncertainty of variable inputs and outputs of hydrological and streamflow simulators used for water level and discharge forecasting in rivers (see here and here). The German Office of Environment has the mandate of operational flow forecasting on the river Rhine waterway. The goal of operational forecasting is to predict low and high discharges/water levels several days ahead along the Rhine and to inform shipping authorities to ensure safe inland water transport. The agency is committed to continuously improve its forecasting capabilities by adopting cutting-edge uncertainty processing methods and to assess the uncertainty of their forecasts. In this project we compare results of the currently used EMOS processor of uncertainty (see here) with the MCP to identify potential advantages and disadvantages of the two methods. The project will deliver a software that will be used to preprocess temperature and precipitation inputs into the River Rhine Model and to post-process streamflow predictions.

Project: "GeCC-LAG-ENSEMBLES: a Generalized Calibration and Combination approach to mix in an optimum way lagged multi-model ensemble forecasts"

Funding Period: 2022-2026

Funding agency: German Science Foundation 

Status: Active 

The project aims at develoing a Generalized Calibration and Combination (GeCC) methodology and apply it to combine in an optimum way lagged ensembles generated by different weather forecasting models. This technique could be used to merge ECMWF medium-range, sub-seasonal and seasonal ensembles and to generate skilful operational products that benefit from the increased ensemble membership for the forecast times, for which more than a single ensemble is available. The generalized combination technique involves linear weights that are determined on the basis of two Bayesian methods: Bayesian Model Averaging (BMA) and the correlation-based Model Conditional Processor (MCP) method. The proposed approach is expected to enhance the skill of the forecasts, especially in the first part of the forecasting horizon at the sub-seasonal and seasonal scale. If successful, the method could lead the way toward smarter and computationally more efficient ways of constructing meteorological reforecasts, that currently absorb considerable computational resources of operational forecasting services. The proposed method also provides calibrated probabilistic forecasts expressed in terms of predictive mean and variance estimates for meteorological variables over the spatial model grid. The predictive statistics are calibrated against historical records. The proposed technique could be used operationally to merge lagged global and limited-area ensembles, generated by different centres, for example by ECMWF and Deutscher Wetterdienst (DWD). Finally, another important aim of the method is to investigate how the forecasting products perform in the tails of the distributions, which represent the meteorological extremes. 

Dummy

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Project: "Precipitation patterns, snow and glacier response in High Mountain Asia and their variability on sub-decadal time scales (PRIME)"

Funding Period: 2017-2023

Funding agency: German Science Foundation

Status: Concluded

We studied the impact of uncertain precipitation estimates on simulated streamflows for the poorly gauged Yarlung Tsangpo basin (YTB), High Mountain Asia (HMA). A process-based hydrological model at 0.5 km resolution is driven by an ensemble of precipitation estimation products (PEPs), including analyzed ground observations, high-resolution precipitation estimates, climate data records and reanalyses over the 2008-2015 control period. The model is then forced retrospectively from 1983 onward to obtain seamless discharge estimates till 2007, a period for which there is very sparse flow data coverage. Whereas temperature forcing is considered deterministic, precipitation is sampled from the predictive distribution, which is obtained through processing PEPs by means of a probabilisitc processor of uncertainty. The employed Bayesian processor combines the PEPs and outputs the predictive densities of daily precipitation depth accumulation as well as the probability of precipitation occurrence, from which random precipitation fields for probabilistic model forcing are sampled. The predictive density of precipitation is conditional on the precipitation estimation predictors that are bias-corrected and variance adjusted. For the selected HMA study site, discharges simulated from reanalysis and climate data records score lowest against observations at three flow gauging points, whereas high-resolution satellite estimates perform better, but are still outperformed by precipitation fields obtained from analyzed observed precipitation and merged products, which were corrected against ground observations. The applied methodology indicates how missing flows for poorly gauged sites can be retrieved and is further extendable to hydrological projections of climate.

Project: " Downscaling of precipitation: development, calibration and validation of a probabilisitc Bayesian approach (BSCALE)"

Funding Period: 2017-2022

Funding agency: German Science Foundation

Status: Concluded 

Downscaling of atmospheric model output is necessary to map variables from low-resolution spatial scales of observation or model prediction down to local scales, at which variables are needed for a wide range of applications, including data gap filling, hydrological or glaciological predictions, climate prognosis, irrigation or energy forecasting. Statistical downscaling is performed by seeking stochastic relationships between large-scale observed indicators and/or model output, serving as predictors, and a local-scale predictand. The underlying transformations are usually linear regressions, or more general non-linear transformations, such as quantile matching. In both cases, stationary homoscedastic relationships between stochastic variables are assumed, which correctly map the conditional mean across the transformation, but not necessarily the tails of the distributions, which characterize extreme meteorological events.

In this project we developed a probabilistic downscaling approach for precipitation, implemented as a Bayesian conditional processor, which supports non-linear transformations between meso-scale observations and model predictions with local variables, whereby stochastic dependency relationships are modelled in the Gaussian space. The procedure allows using multiple predictors over a spatial window, and can be extended to include multiple source models. By using Multivariate Truncated Normal Distributions (MTND), heteroscedastic dependency structures between transformed variables can be modelled in the Gaussian space, then marginalized analytically with respect to predictors and back-transformed into the original space. The downscaling of the Bayesian conditional estimate of precipitation from the meso-scale to the local scale was performed with a non-Markovian non-stationary stochastic weather generator. The processor was validated retrospectively with daily precipitation data in Switzerland.